MEGA: Machine Learning-Enhanced Graph Analytics for Infodemic Risk Management.

Area of Science:

• Network analysis
• Computational social science
• Machine learning

Background:

• The COVID-19 pandemic fueled a significant infodemic, characterized by the rapid spread of misinformation across online social networks.
• Network analysis is vital for understanding and mitigating infodemic risks through statistical and computational methods on large-scale network data.

Purpose of the Study:

• To propose Machine Learning-Enhanced Graph Analytics (MEGA), a novel framework designed to improve the efficiency and performance of learning on massive graphs.
• To apply the MEGA framework to infodemic risk analysis, specifically for detecting spambots and identifying influential spreaders of misinformation.

Main Methods:

• The MEGA framework integrates feature engineering with graph neural networks to process and analyze large network datasets.
• Infodemic risk analysis within MEGA involves detecting spambots using triangle motif counts and identifying key spreaders via distance centrality computations.

Main Results:

• Evaluation on a COVID-19 Twitter dataset demonstrated MEGA's superior computational efficiency compared to existing methods.
• The framework achieved high classification accuracy in detecting misinformation-related network behaviors.

Conclusions:

• The MEGA framework offers an effective and efficient solution for analyzing massive graphs in the context of infodemic risk assessment.
• This approach enhances the capabilities of fact-checking and misinformation detection in online social networks during public health crises.